jminc/resume-matching-dataset-v2

📄 Dataset Card - Resume Matching Dataset v2

Overview

This dataset is designed for training and evaluating large language models (LLMs) on resume-job matching tasks, specifically in AI and software engineering domains.

• All data samples were generated using GPT-4o-mini.
• The dataset exclusively contains synthetic data — no real resumes, self-introductions, or job postings are used.
• This dataset targets three roles:
  • AI/LLM Developer
  • Frontend Developer
  • Backend Developer

Each sample includes:

• A job post (jobpost)
• A resume (resume)
• A self-introduction (selfintro)
• Grade evaluations (resume_grade, selfintro_grade)
• Matching evaluation (evaluation) with total and component scores

---

📂 Structure

Field	Description
jobpost	Synthetic job posting text
resume_grade	Resume quality grade (상/중/하)
selfintro_grade	Self-introduction quality grade (상/중/하)
resume	Synthetic resume text
selfintro	Synthetic self-introduction text
evaluation	GPT-generated evaluation including feedback and reasoning
total_score	Total matching score (0–100)
resume_score	Resume-only matching score (0–50)
selfintro_score	Self-intro-only matching score (0–50)

---

🧪 Example

{
  "jobpost": "React.js 기반 웹 프론트엔드 개발자를 모집합니다. 2년 이상의 경력 필수...",
  "resume_grade": "상",
  "selfintro_grade": "중",
  "resume": "홍길동은 3년간 프론트엔드 개발 경험을 보유하고 있으며...",
  "selfintro": "저는 사용자 경험을 최우선으로 고려하는 프론트엔드 개발자입니다...",
  "evaluation": "이력서와 자기소개서는 직무 요건과 대체로 부합합니다. 총점: 88/100. 강점: React 프로젝트 경험, 팀 프로젝트 협업 경험. 보완점: Next.js 경험 미기재.",
  "total_score": 88,
  "resume_score": 45,
  "selfintro_score": 43
}

📊 Dataset Statistics

• Total Samples: 74,147
• Target Roles: AI/LLM Developer, Frontend Developer, Backend Developer
• Data Source: Fully synthetic (GPT-4o-mini generated)
• Evaluation: GPT-4o-mini scoring and reasoning
• Language: Korean

⚠️ Limitations

• All data is synthetically generated and may not fully capture the nuances of real-world hiring processes.
• This dataset is intended for research and educational purposes only. It should not be used for real-world hiring or personnel decisions.

🏗️ Dataset Creation Notes

• Developed as part of a team project focusing on AI-based career matching systems.
• Data generation, grading, and evaluation were fully automated using GPT-4o-mini.
• No personally identifiable information (PII) or copyrighted material is included.

🔗 Citation If you use this dataset, please cite:

  @misc{resume-matching-dataset-v2,
    title={Resume Matching Dataset v2},
    author={Rezoom Team},
    year={2025},
    howpublished={\url{https://huggingface.co/datasets/jminc/resume-matching-dataset-v2}}
  }